38b455d541
Cosmit pep8
2012-12-25 13:16:05 +01:00
9701ce7fac
remove the random projection option from the 20 newsgroups example
...
- the output will be dense and the models expect sparse input.
- the hyperparameters would need to be re-tuned for the new
dimensionality. f1 scores without re-tuning is misleading
Better have dedicated examples for random projections.
2012-12-21 14:34:19 +01:00
8eb855446a
make it possible to use random projection on the 20 newsgroups classification example
2012-12-21 14:34:18 +01:00
8788713c3b
Update document classification example.
2012-11-05 22:38:11 +09:00
8c9cc855d6
Can afford better precision in news20 example.
2012-09-29 02:16:05 +09:00
107abb6f3e
Add lsqr solver.
2012-09-29 02:16:05 +09:00
9d5936ee87
Revert "BUG rm RidgeClassifier from 20newsgroups"
...
This reverts commit 179eabcba4
2012-09-19 14:08:52 +02:00
179eabcba4
BUG rm RidgeClassifier from 20newsgroups
...
Current implementation has quadratic space complexity in terms of X.
Should use scipy.sparse.linalg.lsqr for linear space version (@mblondel).
2012-09-15 16:05:27 +02:00
16a5719558
BUG fix broken top-10 features printing in text clf example
2012-09-11 10:41:00 +02:00
e55e2c568d
Normalize training and test times.
2012-08-21 15:17:44 +09:00
8043efbc06
Add all_categories option.
2012-08-21 14:42:25 +09:00
833fc05e09
COSMIT simplify a bit examples
2012-05-05 21:57:05 +02:00
15455263b7
API : change back default C to 1. explicitely and epsilon 0.1
2012-05-05 21:57:04 +02:00
56b3167771
COSMIT pep8 document classification example
2012-04-24 21:53:00 +02:00
91a7b65557
DOC small fixes to NearestCentroid classifier
...
Point out connection to Rocchio classifier from NLP/IR.
2012-03-27 18:48:38 +02:00
d213168bef
metric fixed in tests
2012-03-27 19:48:45 +11:00
f5de79a91c
Removed unneeded numpy.array call in test
...
Added NearestCentroid to 20 newsgroups example
2012-03-27 19:40:50 +11:00
c0541fa03b
Renamed Vectorizer to TfidfVectorizer + deprecation warning
2012-03-07 06:48:56 -08:00
b7a77e8962
demonstrate stop words in example (+ slighly faster convergence)
2012-03-05 10:58:11 +01:00
38d4332293
factorize feature names array
2012-03-03 19:19:33 +01:00
631705c49f
merge master
2012-03-03 19:02:45 +01:00
0187838ebe
DOC document classification plot
2012-03-03 17:52:11 +01:00
11d6116de5
missing C re-scaling in example
2012-03-03 17:26:56 +01:00
882dd38faf
missing C re-scaling in example
2012-03-03 17:26:20 +01:00
a510d17a05
Flatten the feature extraction API
2012-03-03 17:20:32 +01:00
5e92beb694
ENH sublinear tf scaling in TfidfTransformer
2012-02-04 16:51:40 +01:00
bb1be4e0ef
Add Perceptron to document classification example.
2012-01-29 02:42:01 +09:00
d5f02dae3d
Merge branch 'master' into merge-linearsvcs
...
Conflicts:
examples/document_classification_20newsgroups.py
2012-01-17 09:21:33 +01:00
9b0fbab0d1
ENH merge dense/sparse LinearSVC, part 3: deprecate sparse.LinearSVC
2012-01-16 18:27:14 +01:00
36c2dcb804
Updat examples.
2012-01-14 22:40:29 +09:00
e5628f79c6
peping8 examples
2011-12-19 18:16:51 +01:00
7375d50de1
DOC typo
2011-11-07 22:01:31 +01:00
bc02942bc3
Nearest Neighbors examples & documentation
2011-09-11 10:45:34 -07:00
b93cde81bd
Documenting a secret feature and fixing bugs in the process.
2011-09-06 19:12:00 +09:00
ddf4b72109
Move project directory from scikits.learn to sklearn
2011-09-02 12:06:57 +02:00
68f27e3790
Revert "Move project directory from scikits.learn to sklearn"
...
This reverts commit fd0d3b879d
2011-09-02 12:03:18 +02:00
fd0d3b879d
Move project directory from scikits.learn to sklearn
2011-09-02 11:38:24 +02:00
5d1b1e9836
API+ENH: load data by default in mlcomp and 20news
...
Conflicts:
scikits/learn/datasets/__init__.py
2011-07-30 14:51:06 +02:00
130b84d724
Rename f_chi2 to chi2
2011-07-23 00:30:21 +02:00
486044bc8c
Demo chi2 feature selection on document classification
...
Also update the example script to use optparse.
2011-07-23 00:30:21 +02:00
bd72672cc4
ENH show top 10 terms per category in document classifier example
2011-07-15 14:26:40 +02:00
514c4f3384
Use four categories instead of all in doc. class. example
...
Accidentally set to all in the course of fixing TfidfTransformer.
2011-07-04 15:30:45 +02:00
09d33353e0
Merge pull request #228 from larsmans/tfidf
...
TfidfTransformer: user-selectable norm
2011-07-04 05:15:41 -07:00
48a3cd1d4c
Better performance of Ax=b solver when b is 2d and A is sparse, and add
...
new tol parameter.
2011-07-01 16:56:07 +09:00
d673f828aa
document classification example doesn't demo only linear classifiers anymore
2011-06-29 13:59:05 +02:00
1cd62ba2ad
Use 10 neighbors in k-NN document classification
...
Boosts F1 from 0.435 to 0.531.
Still a very bad score, while Manning, Raghavan and Schuetze
(Introduction to Information Retrieval) report much higher scores
for similar tasks, with a euclidean distance metric. TODO: find
out why our scores are so low.
2011-06-26 15:35:06 +02:00
1ff5944260
Adapt kNN classifier to sparse input
...
Includes demo on document classification task (with poor performance,
but very acceptable speed).
2011-06-22 11:57:49 +02:00
7880af76c9
Optimize BernoulliNB + improve docstring + add to doc-class example
...
Still two orders of magnitude slower than MultinomialNB because of loop in
_joint_log_likelihood.
2011-06-06 11:37:41 +02:00
d7499b864a
fix message in document classification example
2011-06-06 10:55:32 +02:00
bd8c8febb7
reformulate MultinomialNB as linear classifier
2011-05-23 22:31:15 +02:00
Andreas Mueller